Observational data: Understanding the real MS world
- 11 July 2016
- journal article
- review article
- Published by SAGE Publications in Multiple Sclerosis Journal
- Vol. 22 (13), 1642-1648
- https://doi.org/10.1177/1352458516653667
Abstract
Randomised clinical trials are the primary source of evidence, guiding the use of disease-modifying drugs in multiple sclerosis. However, the spectrum of questions that can be answered in the trial setting is relatively narrow. ‘Real-world’ observational data analysis has always been the major source of evidence for epidemiology, aetiology, outcomes and prognostics, but is now also increasingly used to study treatment effectiveness. While analyses of observational cohorts typically offer superior power, generalisability and duration of follow-up relative to prospective randomised trials, they are also subject to multiple biases. It is the role of researchers to mitigate bias and to ensure the results of observational studies are robust and valid. In this review of observational data research, we provide an overview of the inherent biases, the available mitigation strategies, and the state and direction of contemporary treatment outcomes research. The review will help clinicians critically appraise published results of observational studies.Keywords
This publication has 33 references indexed in Scilit:
- Defining reliable disability outcomes in multiple sclerosisBrain, 2015
- Can we measure long-term treatment effects in multiple sclerosis?Nature Reviews Neurology, 2014
- Methodologies for data quality assessment and improvementACM Computing Surveys, 2009
- SOME PRACTICAL GUIDANCE FOR THE IMPLEMENTATION OF PROPENSITY SCORE MATCHINGJournal of Economic Surveys, 2008
- Sensitivity analysis after multiple imputation under missing at random: a weighting approachStatistical Methods in Medical Research, 2007
- Sensitivity Analysis in Observational StudiesPublished by Wiley ,2005
- 6. Matching with Multiple Controls to Estimate Treatment Effects in Observational StudiesSociological Methodology, 1997
- Indicator and Stratification Methods for Missing Explanatory Variables in Multiple Linear RegressionJournal of the American Statistical Association, 1996
- Inter‐ and intrarater scoring agreement using grades 1.0 to 3.5 of the Kurtzke Expanded Disability Status Scale (EDSS)Neurology, 1992
- Reducing Bias in Observational Studies Using Subclassification on the Propensity ScoreJournal of the American Statistical Association, 1984